Quinnipiac finds McAuliffe ahead in Virginia, but narrowly. Mark Mellman writes a must-read on how "likely voters" miss likely electorates. And in Communist Russia, poll disapproves of you. This is HuffPollster for Wednesday, September 18, 2013.
QUINNIPIAC POLL GIVES MCAULIFFE NARROW EDGE OVER CUCCINELLI - From the release: “With 7 percent of likely voters, Robert Sarvis, the Libertarian candidate in the too-close-to call Virginia governor's race, could hold the key to victory for Democrat Terry McAuliffe, who has 44 percent of likely voters, or Republican State Attorney General Ken Cuccinelli, who has 41 percent, according to a Quinnipiac University poll released today. Democrat McAuliffe gets a split 38 - 38 percent favorability rating, the independent Quinnipiac University poll finds, compared to Republican Cuccinelli's negative 34 - 51 percent rating. For Sarvis, 85 percent don't know enough about him to form an opinion.” [Quinnipiac]
One more VA poll tomorrow Roanoke College, the only pollster since June to put Cuccinelli ahead, says it will release a new survey Thursday morning.
MELLMAN: SAMPLE LIKELY ELECTORATE, NOT 'LIKELY VOTERS - In a column, Democratic pollster Mark Mellman argues that pollsters should care less screening for "likely voters" and more about sampling the "likely electorate." As an example, he points to the New York City comptroller's race between his client, Scott Stringer and Eliot Spitzer: "A New York Times/Siena poll had Spitzer ahead by 15 percentage points less than two weeks before the election, while a slightly earlier NBC 4/Wall Street Journal/Marist poll gave the former governor an 18-point lead. It sounds plausible until you realize that our polling had Stringer ahead the entire time and our final track was just about half a point off the final margin. Why the difference? We focused on the likely electorate, Siena and Marist on likely voters. Our likely electorate was about 25 percent of registered Democrats. The actual turnout was about 23 percent. Siena was looking for likely voters using a carefully constructed model based on responses to 5 questions. However, their likely voters constituted 67 percent of registered Democrats, three times the number that actually turned out. Marist’s likely voters were 59 percent of Democrats, also far more than voted. Siena and Marist were talking to a lot of people who said they were registered Democrats (but may not have been) and claimed they were likely to vote (but didn’t). Those folks are quite likely to be different from the 23 percent who actually voted. They are less politically engaged and less likely to have known Stringer." [The Hill]
Twitter reaction:
-Marist Institute Director Lee Miringoff: "Marist final poll w/ low turnout model had Stringer up by 1%, w/i MoE of final result." [@LeeMiringoff]
-AP Polling Director Jennifer Agiesta: Mellman's column is "one way to assess LVs that's often overlooked [the size-of-the-electorate angle]...Good research is expensive. But even if RDD can't get to 25%, I know it can do better than a 44-pt overestimate." [@JennAgiesta here and here]
-Marist Institute Senior Analyst Natalie Jackson: Mellman is "also comparing polls 2 weeks out to actual results. Even 4 days prior is difficult." [@NatalieMJ84]
-Political scientist Michael McDonald: "@MysteryPollster: Does @MarkMellman disclose how he arrives at "likely electorate"? Was crux of @ppppolls #nerdfight." [@ElectProject]
-HuffPollster replies to McDonald: "Fair point. No, not specifics. My understanding is that it involve lists & vote history. Mark?" [@MysteryPollster]
-Mellman replies to us: "Yep" [@MarkMellman]
-Wall Street Journal reporter Neil King: "But @MarkMellman doesn't tell us how he got to the 'likely electorate.' That's the secret sauce." [@NKingofDC]
-Mellman replies to King: "Flks gotta pay 4 somethin." [@MarkMellman]
So what is the secret sauce? - In the twitter exchange above, Mellman confirms that in New York City, they drew samples from lists of registered voters that include "vote history" -- the record of whether individual voters cast ballots in previous elections. Via email, HuffPollster asked Mellman if he could put aside his concerns about sharing proprietary details and at least explain in general terms how he focuses on the "likely electorate." His response: "In short... [we] use voter history and other data to model the likely electorate based on similar past elections." And what would he say to those who wonder if, as Nate Cohn alleges of PPP, Mellman is weighting data in an ad-hoc way to get results that feel "right?" His answer: "This modeling is all done before the fact, not after we get results....this is about designing the sample, not about weighting after a survey is completed.....and the modeling is based largely on actual vote history....though it is not a simple voted in last two elections or whatever." [For a primer on samples drawn from lists vs. random digits ("RDD"), see this previous HuffPollster report]
GEORGETOWN GETS $100M TO LAUNCH POLICY SCHOOL, 'MASSIVE DATA' INSTITUTE - Nick Anderson: "A former owner of the Los Angeles Dodgers is giving Georgetown University $100 million to found a school of public policy that will bear his family name...The McCourt School of Public Policy, to be launched in October, will join a host of schools with a similar mission at other universities...Georgetown’s version will focus on analysis of massive data files emerging from a plethora of government agencies and other sources in the 21st century, seeking to shed light on education, health, poverty and other subjects...The McCourt gift will fund an expansion of faculty, scholarships for top students, a Center for Global Politics and Policy, and a Massive Data Institute. ...Georgetown Provost Robert Groves, a former director of the U.S. Census Bureau, said the data initiative will capitalize on the explosive growth of quantitative public information through such sites as www.data.gov, to help frame policy issues and train new generations of government leaders. A critical staffing issue for many agencies, Groves said, “is just having the talent that can navigate these files.” [WaPost, see also Georgetown release]
DE BLASIO HAS BIG EARLY LEAD - Michael Howard Saul: “Bill de Blasio leads Republican Joe Lhota among likely voters by 43 percentage points in the race for New York City mayor, an early jolt of positive news for Democrats seeking to recapture City Hall for the first time in 20 years, a Wall Street Journal-NBC 4 New York-Marist poll released Tuesday showed. Mr. de Blasio, the city's public advocate, is outpacing Mr. Lhota, a former deputy mayor under Rudy Giuliani, 65% to 22% among likely voters, including those who are undecided but leaning toward a candidate. Adolfo Carrión Jr., a former Bronx borough president running on the Independence Party line, had 3%.” [WSJ, Marist crosstabs]
HUFFPOLLSTER VIA EMAIL! - You can receive this daily update every weekday via email! Just enter your email address in the box on the upper right corner of this page, and click "sign up." That's all there is to it (and you can unsubscribe anytime).
WEDNESDAY'S OUTLIERS' - Links to more news at the intersection of polling, politics and political data:
-Republicans sour on Ben Bernanke. [Pew Research]
-Support for Occupy Wall Street half of what it was in 2011. [YouGov]
--26 percent of Americans, "disproportionately" conservatives and Republicans, oppose raising the debt limit even though they also believe not raising it will seriously harm the economy. [ABC News]
-Greenberg and Carville release a survey segmentation: "mapping the Republican brain." [Democracy Corps]
-David Hill wants to highlight the "nonpartisan" reactions to Syria and the gun show loophole. [The Hill]
-Harry Enten sees peril for Democrats in Obama's falling approval rating. [The Guardian]
-The new CensusReporter tool, still in beta, is "usable and clearly on its way to an improvement over the painful default." [Flowing Data]
-The Pew State and Consumer Initiatives has posted map-based timeline showing foreign-born residents as a percentage of state population. [Pew Research via @FactTank]
-“For First Time, Americans' Views of Russia Turn Negative,” according to a Gallup poll of people who apparently slept through the Cold War. [Gallup]